A reproducibility-audit framework for generalizable versus dataset- specific molecular transition boundaries in Alzheimer’s disease
Listed in
This article is not in any list yet, why not save it to one of your lists.Abstract
Molecular staging of Alzheimer’s disease (AD) increasingly defines transition boundaries along single-cell pseudo-progression trajectories, yet whether such boundaries reproduce across brain regions, cohorts and molecular modalities is rarely tested. We present a permutation-controlled audit that combines nine boundary-detection algorithms with a fixed marker panel and four orthogonal reproducibility axes—algorithmic consensus, region, cohort and modality. On synthetic data with planted ground-truth boundaries the audit reaches 100% sensitivity and 94% specificity, rejecting four distinct artefact classes each by a different axis. Applied to the Seattle Alzheimer’s Disease Brain Cell Atlas middle temporal gyrus, it localizes a transition that is robust across algorithms and recovered in most cell types but does not generalize: its leading marker is attenuated or absent in prefrontal cortex, entorhinal cortex and cerebrospinal fluid, and an apparent cross-region conservation of glial metabolic genes proves to be a global-expression offset rather than a shared program. The same audit nonetheless certifies an externally validated marker (astrocytic PTGDS) as reproducible across regions and modalities, showing that it separates generalizable anchors from dataset-specific ones rather than rejecting all signals. We provide this four-axis audit as a transferable, code-available standard to apply before a trajectory boundary is read as a biological stage, in AD and other progressive proteinopathies.
Highlights
-
A four-axis audit tests whether AD molecular-staging boundaries reproduce
-
On synthetic ground truth it reaches 100% sensitivity and 94% specificity
-
A robust MTG transition fails to generalize across regions, cohorts and CSF
-
Apparent cross-region metabolic conservation is a global-expression artefact
In brief
Kim et al. develop a permutation-controlled, multi-algorithm audit that tests whether molecular transition boundaries in Alzheimer’s disease reproduce across regions, cohorts and modalities. Validated on synthetic ground truth, the audit separates a generalizable anchor (PTGDS) from a dataset-specific signal and a global-expression artefact, providing a reusable control for trajectory-based molecular staging.
Motivation
Single-cell atlases increasingly place disease stages at molecular transition boundaries along pseudo-progression trajectories, and the marker that anchors such a boundary is often read as biologically meaningful. Yet a boundary identified in one atlas need not recur in another region, cohort or molecular modality, and a large single-dataset effect can reflect global transcriptional amplitude rather than a specific program. No standard procedure exists to decide whether a trajectory boundary generalizes before it is interpreted. We therefore built a transferable, permutation-controlled audit with measurable operating characteristics that makes this decision explicit, and asked what it implies for one widely used atlas.
